Environmental impacts from livestock production with different animal welfare potentials-a literature review

(1)

SIK report 844

Environmental impacts from livestock

production with different animal welfare

potentials – a literature review

Christel Cederberg Maria Berglund Jenny Gustavsson Magdalena Wallman

(2)

(3)

SIK report 844

Environmental impacts from livestock production with

different animal welfare potentials – a literature review

Christel Cederberg Maria Berglund Jenny Gustavsson Magdalena Wallman 2012 ISBN: 978-91-7290-314-2

(4)

(5)

SUMMARY

There is a rising awareness of the many negative environmental impacts related to the rapidly growing global production and consumption of animal products, such as land degradation, biodiversity losses, emissions of nutrients (most importantly nitrogen and phosphorus) and greenhouse gases. Life Cycle Assessment (LCA), and, more recently, Carbon Footprinting (CF), have become the major methods to assess the environmental performance of livestock production, and today there is access to relatively

sophisticated and standardised methods for Carbon Footprinting. Results from LCA/CF research and consultancy work are used in the industry, by policy makers and also, recently, by consumers.

Animal welfare is not included in current LCA or CF analyses of food production systems. The lack of data on animal welfare which is comparable with environmental LCA data means that there can be inadequate knowledge of how sustainable animal production systems can and should develop. This knowledge gap is the background for this study which was commissioned by Compassion in World Farming (CIWF).

Method

This study is based on a thorough literature review of scientific reports on the

environmental impacts of milk, meat and egg production. The reports reviewed were mainly LCA and CF studies. Peer reviewed papers and recent, high quality reports that compare and present results from differing production systems were examined in more detail.

There is limited or no information about animal welfare in the studies analysed. One way to deal with this limitation is to consider grazing and/or organic farming systems and different types of free-range systems as a proxy for systems with a greater potential for higher farm animal welfare.

Results and limitations

Of the studies examined, all include emissions of greenhouse gases from animal production (although emissions due to land use and land use change are seldom

included) and most of the studies present figures on energy use, land use, eutrophication and acidification. Few studies include use of pesticides or toxicity assessments,

phosphorous use or other resource depletion. Assessment of biodiversity is included in only one study, whilst land quality and medicine use are not included in any of the studies reviewed. Land use is quantified in a number of studies, but they do not include any classification of the quality of land. This makes it hard to determine the amount of prime arable land needed versus the amount of upland or marginal land used in farming systems that have the potential for higher or lower animal welfare. Some indicators are given as the sum of total pollutants (e.g. eutrophying and acidifying substances) but their distribution over time and space is not measured. Also, the sensitivity of the ecosystems where those pollutants are assumed to be deposited is not considered. Therefore, those indicators do not reflect actual impact on the environment.

There is a large range in the number of studies for different livestock products, with 30 studies covering dairy and only eight studies covering eggs – two of which partly share the same data source. Many published LCA/CF-studies are based on farm data

(especially for dairy production), while some are limited to modelling farming systems. Which approach provides the most reliable results is determined by the quality of the farm inventory data and the quality of the models. Most of the reviewed studies are

(6)

based on farms in industrialised nations with an emphasis of Western Europe. Studies based on modelled data have a tendency to produce results which were different to the larger body of studies based on farm data. The use of statistical analyses in all except the dairy studies was limited.

An assessment of the availability and quality of the data and literature available for the environmental impacts of livestock systems provides key insights:

Some livestock products are more frequently studied, such as dairy, while others are poorly covered in LCA/CF-studies, including pigs, chicken meat and eggs.

Some environmental impacts are often studied, such as emissions of

greenhouse gases (although seldom including emissions due to land use and land use changes), while others are poorly covered, e.g. biodiversity.

Land use is only presented quantitatively and there is a lack of information on the quality of the land needed for different livestock production systems.

The measure of some environmental indicators is not necessarily equivalent to an environmental impact, for example the sensitivity of the ecosystems where eutrophying and acidifying pollutants are assumed to be deposited is not considered.

Lack of consistent methodologies in published LCA/CF studies, e.g. variations in how data are acquired (real farm data versus farm modelling) or method for co-product handling makes it difficult to compare different livestock production systems from different studies.

LCA/CF-studies are more frequently available for animal production in Western Europe than for livestock farming in the global south.

Dairy

Dairy is the most frequently studied production category (30 studies in total), nine of which include analyses of farming systems with differing animal welfare potentials. Review of these nine studies finds that differences in environmental impacts of milk between different production systems in industrialised countries, and Western Europe in particular, are well documented through LCA studies. Our analysis shows that dairy production systems with the potential for higher animal welfare (mostly organic) have lower energy and phosphorus use and substantially lower pesticide use than current standard conventional systems. Overall land use is higher in organic dairy systems. For climate change, eutrophication and acidification there were no consistent results

between studies, indicating that no one system has an environmental advantage over the other.

Beef

Environmental assessments of beef were found in 21 published studies, of which seven were analysed further. Statistical analyses are included in less than half of these studies. Most of the studies come from Europe and are mainly based on modelled data; only a few include data from commercial farms. This presents difficulties in assessing the data, as beef is produced in a wider range of farming systems than is the case for milk in the industrialised world. Generally, the results of the studies for beef are also more difficult to interpret than those for dairy, since the methodology is less clear in beef LCAs. There

(7)

are clear overlaps between milk and beef production as over 50% of the world´s beef is produced from calves bred in dairy systems and meat from culled cows. In general, dairy cows produce a calf each year, to ensure continued milk supply. Calves are either raised as dairy cow replacements (heifers), killed at birth (mostly unwanted males), or raised for beef. Additionally, dedicated beef herds are kept for meat production. The reviewed LCA/CF-studies for beef do not always make it clear where the boundary lies between dairy and beef systems and how carbon emissions or other impacts are

attributed. One study demonstrates that a reduction in greenhouse gas emissions from the beef sector can be achieved by producing beef from dairy calves (integrating the dairy and beef sectors), even where the bull is a dairy bull. This would reduce the need for beef herds and so reduce overall greenhouse gas emissions. The use of robust dual purpose dairy breeds may also be consistent with higher animal welfare and may be more suitable for beef production.

With the limitations of the data in mind, the seven beef studies examined in depth illustrated a wide range in the production of greenhouse gas emissions between and within studies for different farming systems. Emissions of greenhouse gases are similar or higher for production systems with higher animal welfare. Eutrophying and/or acidifying substances are included in four studies, and according to these, emissions are higher for production systems with higher animal welfare. Land use (included in only two studies) is substantially higher for the extensive/organic systems, but land quality is not assessed. Pesticide use is significantly lower in organic systems than in conventional systems.

Pig production

Environmental assessment of pig production systems was found in 19 published studies, of which only four compare systems with different animal welfare potentials and these were selected for further analysis. The four studies analyse mixed-farming systems in Western Europe, are based on modelled data and do not include statistical analysis. This imposes considerable limitations on the analysis of the results and any conclusions drawn. From the four studies considered, land use was lower in conventional systems whilst pesticide use was lower in organic systems. There were no consistent results for energy use or climate change parameters.

Poultry production

Environmental assessments of chicken meat and egg production systems were found in nine and eight studies, respectively. Three studies in each sector included comparisons of systems with the potential for higher or lower animal welfare and were selected for a more in-depth analysis. However, two of the studies for chicken meat and eggs partially use the same input data, which further limits the data pool. Only one of the selected poultry studies (egg) is based on inventory data from commercial farms and this study is also the only one including a statistical analysis. As with beef and pork, this imposes considerable limitations on the analysis of the results.

There are too few studies and data sets to draw robust conclusions about the relative environmental impacts of conventional poultry farming systems and more extensive systems. Given the small sample size and paucity of farm-based data in poultry studies, there is a great need for more studies comparing the environmental impacts of different production systems before drawing robust conclusions on the differences in

(8)

General outcomes

Published LCA/CF-studies that are suitable for in-depth analysis to compare the environmental performance of different livestock production systems are too limited to draw robust conclusions in many cases. No, or very little, information on animal welfare is available in the studies examining the environmental impacts of animal farming. Also, there is no consensus methodology currently available to assess animal welfare which is comparable to the environmental LCA methodology.

Using the data available, no consistent differences in the environmental performance of farming systems with differing animal welfare potentials were found for most

environmental indicators with the exception of land use and pesticide use/toxicity. Organic and more extensive systems require more total land area per unit of product, but there is very little data on the use of arable or marginal land per product and the impact of the land use. The second consistent finding is that organic systems consistently use less pesticide than conventional farming systems.

Need for future research

More research is needed on the interaction between milk and beef and how to make best use of the two sectors when developing production systems with lower environmental impact and higher animal welfare. An interesting scenario to analyse in more depth is the development of a dual-purpose dairy system, suitable also for meat production. This could allow a reduction in the number of beef breeding cows which require large land areas and produce meat with a very high carbon footprint.

It would be beneficial for new impact assessment studies to be more site- and area-specific and to measure the environmental impact of farm activities rather than

indicators from inputs and outputs. More detailed analysis of nitrate would allow greater understanding of the impact on the environment and drinking water quality.

More complete LCA-studies of industrial production of pig and poultry are required. Livestock density may vary considerably between production systems. Hence, local and regional effects of high livestock density on small areas having regard to groundwater contamination, eutrophication and acidification should be included in such analyses. In addition, assessments should be based on validated data, e.g. representative data from commercial production units. Since industrial pig and poultry production are growing fast in developing countries, studies are needed from these regions.

The strong focus on climate change in recent years has been an important driver for the development of consensus around methods for calculating carbon footprints of products. In order to have more complete information on the impacts of animal farming, there is an urgent need for a similar development of accepted methods which include

biodiversity impact, land quality issues, medicine use and animal welfare issues, in order to increase information on sustainability in animal production systems. Also, the issues of land use change and land management are very important. GHG emissions caused by soil carbon changes due to land use management are poorly reflected in existing studies and there is a lack of consensus around methodology. GHG emissions due to direct and indirect land use changes are important and there is quite extensive research underway for developing mutual models and methodology.

The growing use of veterinary medicines over the last 30 years, accompanied by growing problems of the development of resistant bacterial strains and emissions of medicines into the environment, is an important research field where more knowledge is needed generally.

(9)

The environmental and animal welfare impact of animal production also rests on overall levels of production and consumption and the efficiency of the supply chain. Other ways of reducing the environmental impacts of animal production systems are reducing production and consumption of animal products and reducing waste along the supply chain. More research on methods to reduce losses which have an impact on both

environment and animal welfare, from early culling and death of animals due to disease, lameness or injury, would be useful.

ACKNOWLEDGEMENTS

This research has been commissioned by Compassion in World Farming (CIWF), Godalming, Surrey, UK, with funding from CIWF, The Tubney Charitable Trust, the World Society for the Protection of Animals (WSPA) and the Swedish Institute for Food and Biotechnology (SIK).

Key words

Environmental impact, livestock, animal welfare, LCA, Carbon Footprint, milk, meat, eggs.

(10)

(11)

1. INTRODUCTION

During recent years there has been a rising awareness of the many environmental impacts caused by the rapidly growing global production and consumption of animal products. According to the FAO report “Livestock´s Long Shadow”, the global livestock sector is one of the top two or three most significant contributors to some of today‟s most important environmental problems, at every level, from local to global. According to this FAO study, the production of meat, milk and eggs is having a major impact on land degradation, climate change, air pollution, water shortage and loss of biodiversity.

There is a growing interest and use of the environmental analysis method Life Cycle Assessment (LCA) which is now the major method used in the research of the

environmental impact of livestock products. The recent focus on global warming has led to LCA being further developed as a tool for carbon footprint accounting, i.e. a method for estimating a product´s greenhouse gas emissions throughout its life-cycle. Today we have access to relatively sophisticated methods to calculate the environmental

performance of meat, milk and eggs, and results from research and consultancy work in this area are used in the industry, by policy makers and, more recently, by consumers.

Purpose of this study

While there are standardised methods to assess the environmental burden of animal products, there are far fewer tools to assess and score animal welfare potential in livestock production systems. The lack of such tools means less information on animal welfare issues as opposed to environmental performance and how the issues of

environmental performance and animal welfare relate to each other. This has resulted in an inadequate knowledge as to how a sustainable animal production system can and should develop, considering these two important aspects of production. LCA and carbon footprinting assess environmental impacts but do not include any aspects of animal welfare. This knowledge gap is the background for this study which has been

commissioned by Compassion in World Farming (CIWF) and funded by CIWF, The Tubney Charitable Trust and the World Society of the Protection of Animals (WSPA). The study has several purposes:

- Describe and analyse the most common methods and tools used for assessing the environmental impact of livestock production;

- Investigate to what extent different livestock production systems can be environmentally assessed (scope of environmental problems, geographical extent etc);

- Analyse differences in environmental impact between livestock production systems, including varying animal welfare standards;

- Identify gaps of knowledge in the area of environmental performance in global livestock production systems in relation to important animal welfare issues.

Method

The method used in this work is a thorough literature review of scientific reports on environmental impacts from animal products with a focus on primary production in agriculture. Studies that compare production systems with distinct animal welfare potential are selected and analysed in more depth and the outcomes of these studies are compared for different impacts. Good animal welfare depends on a number of factors,

(14)

including good management, breeding (i.e. animals fit for their environment) and the conditions in which the animals are reared. Welfare potential describes how good welfare can be part of a farming system, and is determined by genetics and

environment. Since there is limited information about animal welfare in the analysed studies, one way of dealing with this limitation is to consider organic production as a proxy for a system with a higher welfare potential. For dairy and beef, pasture-based systems, such as the New Zealand dairy model or extensive beef production systems, can also be used as proxies for systems with a higher welfare potential. For pigs and poultry, free-range systems, outdoor reared or certain indoor systems, such as deep-litter systems, can be used as proxies for systems with higher animal welfare potential.

Systems with low welfare potential include close confinement systems such as the battery cage for laying hens and the sow stall for breeding pigs, as they fail to meet animals‟ basic needs for space to move and express natural behaviours. More extensive systems such as free-range and some indoor systems are described as having a higher welfare potential, although good management and stockmanship is required in all systems to meet their full welfare potential.

In LCA and/or carbon footprint (CF) accounting studies, results can be significantly affected by methodology choices of the LCA executant. Examples of important

decisions include: setting system boundaries, co-product handling, and choice of models for estimating emissions of greenhouse gases and reactive nitrogen. Differences

between animal production systems found in different studies can be a consequence of varying methodology choices rather than a reflection of reality. This is the main reason why a number of studies are singled out in which different animal welfare potentials are assessed with uniform methods. Studies for deeper analysis in this report were selected to include those with a standardised and thorough LCA methodology, from scientific papers/reports that are either peer reviewed or include well-documented and transparent methodologies.

Structure of the report

The report starts with a background section in which key environmental impacts connected to livestock production are described and the methodology of LCA and CF, which are closely related to one another, is presented. There is also a description of current gaps of consensual methods, for example for biodiversity impact assessment. The next chapter focuses on a description of two LCA research models developed in the UK and Switzerland and it provides an overview of recently developed tools for carbon footprint accounting to be used at farm level, by farmers and/or advisers.

The literature review is presented in two parts: first, the list of total studies found which is presented by different animal product and includes information on the scope of environmental assessments in the studies. The second part includes the analysis of the selected publications comparing the environmental impact between production systems with different animal welfare potential, where similarities and differences are

considered. In the final discussion, the results from the literature review are examined further in light of the environmental problems generally known to be associated with animal production, along with the limitations in the LCA methodology. Based on the findings from the analysis, conclusions are made, with a focus on identifying the need for more research.

(15)

2. GLOSSARY

Allocation (through partitioning)

Used in LCA to partition the environmental impact of a process or product system between several products that are i) produced in the same system, e.g. milk and beef from dairy cows, or ii) the input to a common process, e.g. various waste products that are landfilled at the same site.

Black carbon “Soot”. Formed through incomplete combustion processes such as the burning of fossil fuels and biomass.

By-product A co-product with substantially lower economic value than the other co-product(s) from the system.

Carbon dioxide equivalents, CO2e

Unit to compare the radiative forcing of greenhouse gases and relate it to carbon dioxide.

Carbon Footprint, CF

Total lifecycle greenhouse gas emissions attributed to provide a specific product or service.

Carcass weight Slaughter weight, meat with bone.

Co-product Any of two or more products (goods or services) coming from the same process or product system. Products are considered co-products if one of them cannot be produced without the other being produced, e.g. milk and beef from dairy cows. Co-products have economic value, in contrast to waste.

Crop land Is here considered as arable lands and permanent crops. Arable lands include land under temporary crops (e.g. cereals), temporary meadows for mowing or pasture, land under market and kitchen gardens and land temporarily fallow. Permanent crops include crops that occupy the land for long periods and need not be replanted after each harvest, such as cocoa, coffee and vines.

Denitrification Microbial reduction of nitrate to nitrogen gas. The process is anaerobic, i.e. occurs in the absence of oxygen.

Enteric fermentation

Digestive process by which carbohydrates are degraded by microorganisms and methane (CH4) is formed as a by-product. CH4 emissions are high from ruminant livestock, e.g. cattle and sheep, since the microbial fermentation

supports these animals with much of the nutrients they need. Eutrophication Process by which a body of water becomes enriched in plant

nutrients (i.e. nitrate or phosphates) that stimulate the growth of aquatic plant life.

Functional unit, FU

Term used in LCA to denote the reference unit to which all inputs and emissions are related. The FU quantifies the service provided by the studied system.

Grazing livestock production systems

Production systems based mostly on grazing and feed from different types of grasslands, often found in marginal areas (also called pastoral systems).

GtC Giga tonne carbon

Life Cycle

Assessment, LCA

Total lifecycle environmental impact attributed to provide a specific product or service. Includes several environmental impact categories, e.g. climate change, eutrophication and acidification, and use of resources, e.g. land, phosphorus and water.

(16)

LULUC Land use and land use change Industrial livestock

production system

Production systems in which less than 10 per cent of the feed is produced within the production unit (so-called “landless” systems), mostly used for poultry and pork production. Mixed animal

production system

Production systems where cropping and livestock rearing are associated in mixed farming and more than 10 per cent of the feed comes from crops on farm.

Nitrification Aerobic, i.e. require oxygen, microbial oxidation of ammonium to nitrate.

Pasture Corresponds to “permanent pastures” as defined by FAO, i.e. land used permanently (five years or more) for herbaceous forage crops, either cultivated or growing wild.

Primary energy Energy form found in nature that has not been subjected to any conversion or transformation processes, e.g. uranium, geothermic energy and biomass.

Secondary energy Energy transformed from of other sources of energy. Some examples are electricity, gasoline and ethanol.

System expansion Adding processes or products (and their related emissions) to a system. Used to make several systems that partly provide different products comparable.

3. Abbreviations

ADI Acceptable daily intake a.i. active ingredient

AU Australia AT Austria BR Brazil CA CF Canada Carbon Footprint CH4 Methane CH Switzerland CO2 Carbon dioxide

CO2e Carbon dioxide equivalents CW Carcass weight

DBC Dairy bull calf system

DK Denmark

DMI Dry matter intake

ECM Energy corrected milk (e.g. milk with 3.4 % protein and 4 % fat content) EF Emission factor

ES Spain

EU27 The 27 member states of the European Union

FAO Food and Agriculture Organization of the United Nations FCM Fat corrected milk (i.e. milk with 4 % fat content)

FN Finland

FR France

FU Functional unit

GHG Greenhouse gas emission(s) HSCW Hot standard carcass weight

IE Ireland

ILUC Indirect land use change

IPCC Intergovernmental Panel on Climate Change ISG Indicator species group

(17)

IT Italy

JP Japan

LCA Life cycle assessment LCI Life cycle inventory LUC Land use change

LULUC Land use and land use change

LULUCF Land use, land use change and forestry

LW Live weight

MS Milk solids Mton Million tonnes

N Nitrogen N2O Nitrous oxide NH3 Ammonia NL Netherlands NO Norway NO3- Nitrate NZ New Zealand P Phosphorus PE Peru PT Portugal

SALCA Swiss Agricultural Life Cycle Assessment SCC Suckler-cow calf system

SE Sweden

SOC Soil organic carbon

UK United Kingdom

US United States

(18)

(19)

4. KEY ENVIRONMENTAL IMPACTS CONNECTED WITH ANIMAL

PRODUCTION

Nutrient-related impacts

The mismanagement of nutrients is one of the important negative environmental impacts of livestock production. Emissions of reactive nitrogen (N) from the animal product chain are significant contributors to eutrophication, acidification and climate change. Phosphorus (P) emissions also contribute to eutrophication. Its high use in agriculture causes rapid depletion of an important non-renewable resource, phosphate rock. Livestock absorb between 15-40% of the N and P ingested in their diet, and consequently the majority of these nutrients in feed ends up in the manure. The high concentration of nutrients in livestock manure makes aspects of management and application of manure essential when assessing the environmental impact of livestock production and analysing mitigation options.

Mismanagement of nutrients in the animal production chain is one reason for the large negative human interference with global N and P cycles. In 2005, over 100 million tonnes (Mton) of N in synthetic fertiliser and leguminous plants were used in global agriculture, but only 17 Mton of N were consumed by humans in the form of crops, dairy and meat products. This highlights the extremely low efficiency in N-use in modern agriculture which has changed for the worse over recent decades; the global nitrogen-use efficiency of cereals decreased from ~80% in 1960 to ~30% in 2000 (Erisman et al., 2008). In addition the P cycle has large deficits according to Cordell et al. (2009), less than 20% of the yearly mined phosphate rock aimed for fertilisers ends up as phosphorus in human food.

Nitrogen

The anthropogenic input of nitrogen to the biosphere has grown substantially since the 1950s due to the need for increased food production. Annually it is estimated at around 140 Mton of N, of which ~85 Mton are industrial fertiliser production, ~33 Mton are biologically fixed in leguminous crops and ~21 Mton are produced in combustion processes and emitted as nitrogen oxides (Erisman et al., 2011). Consequently, around 120 Mton of new reactive nitrogen (synthetic fertilisers and leguminous crops) is used in agriculture every year, see Figure 1.

Figure 1. Yearly anthropogenic input of nitrogen (N) to the biosphere in the early 2000s. 0 20 40 60 80 100 120 140 160 Indust fertiliser prod Biological N-fixation Combustion processes Total M ton N

(20)

Nitrogen excreted from livestock globally in manure is estimated by Beusen et al. (2008) at 112 Mton (range 93-132 Mton) per year, equivalent to the input of new reactive N to agriculture. N in manure is distributed as about two thirds in mixed and landless systems1, 20% in pastoral systems2 and around 15% ending up outside the agricultural system (e.g. manure used as fuel), see Table 1.

Table 1. Global estimate of nitrogen in excreted manure for the year 2000.

Livestock system Manure N, Mton per yr

(range)

Share

Mixed and landless systems 74 (58-92) 0.66

Pastoral systems 22 (15-31) 0.20

Non-agricultural, e.g. fuel 16 (6-29) 0.14

Total 112 1.0

Source: Beusen et al., 2008

Ammonia from livestock manure is a major pathway of losses of reactive N from agriculture. Beusen et al. (2008) estimate total ammonia losses at about 21 (16-27) Mton NH3-N per year, 44% (31-55%) of this comes from animal housing and storage systems, 30% (23-38%) from spreading of manure and 27% (17-37) from grazing animals, see Figure 2. Ammonia emission rates are generally lower for grazing systems compared to mixed/land-less production systems, in which ammonia emissions occur in several steps of the manure handling; in housing, storage and when spreading manure (Beusen et al., 2008).

Figure 2. Distribution of global ammonia losses from livestock.

Nitrogen from agricultural land, mostly leached nitrate, is a major contributor to the eutrophication problem and to nitrate pollution of groundwater which is a recognised risk for its use as drinking water. Excess inputs of N and P into surface waters can lead to algal growth, oxygen deficiencies and fish mortality. Bouwman et al. (2009) estimate losses of N in leaching and erosion in the year 2000 at around 40 Mton N globally. The indicator Nitrogen loading indicator is used for assessing watershed nutrient loads and shows that in Europe, water pollution from nitrogen is mainly the result of livestock production and fertiliser use. In India and southern parts of South America, livestock production is the dominant contributor while in North America and China fertilisers

1

Mixed systems are defined as livestock systems where 10-90% of animals feed is produced at the farm; landless system is defined as systems where <10% of feed is produced at the farm.

2_{Pastoral systems are extensive systems producing beef and milk and cattle are principally grazing all}

year around.

Animal housing and manure storage Manure spreading

(21)

dominate the total N load (Erisman et al., 2011). However, when applying a life cycle-perspective on these findings, considerable share of N-fertilisers are used in feed crops (e.g. 60% of the European cereals are used for feed), thus the leaching from these crops should be attributed to the livestock sector.

Emissions of the greenhouse gas nitrous oxide (N2O) also represent a loss of reactive N in agriculture, but in absolute terms, they account for far smaller loss than those from reactive N as ammonia and nitrate. Denman et al. (2007) estimates that around 2.8 Mton of N yr-1 (1.7-4.8) is lost as N2O-N from agriculture, mainly due to denitrification and nitrification processes in the soil, and nitrogen transformations in manure.

Phosphorus

Mined phosphate deposits are mainly used in food production; as fertilisers (80%) and mineral feed (5%), while the remaining 15% goes to industrial uses, mostly detergents. At current usage levels it is estimated that today´s economically exploitable resource will be depleted within 125 years and total reserves (the “reserve base”) within 340 years (Smit et al., 2009). However, when the expected increase in world population, the projected growth in meat production and consumption and the anticipated increase in biofuel crops in agriculture are all taken into account, exploitable resources will only last for around 75 years, unless improvements are made in the efficiency of use of phosphorus in agricultural systems (Smit et al., 2009).

In the early 2000s, fertiliser use was roughly 15 Mton of P per year in agriculture. As a global average, Cordell et al. (2009) estimate that approximately 15 Mton of P is produced in livestock manure annually of which around half is returned to agriculture and the rest is lost via land-fill, non-arable soils and waters. Phosphorus in human excreta makes up about 3 Mton of P per year, of which only around 10% is returned to agriculture. In Figure 3, the annual P flows in synthetic fertilisers, manure and human excreta are illustrated. It is obvious that livestock manure and human excreta must be better recycled to agriculture in order to reduce the continuous growth in use of fertiliser phosphorus.

Figure 3. Yearly P flows in fertilisers, manure and human excreta. Based on Cordell et al. (2009).

Approximately half of global P fertilisers are used in cereals. This means that animal production is a major driver in the depletion of the phosphorus resource, as worldwide around 30% of cereals are used as livestock feed. In Europe this share is 60%. The important protein source soybean is another P-demanding crop. P-fertilisation has

-10 -5 0 5 10 15 20 P-fertilisers Manure, to agriculture

Manure, lost Human

excreta, to agriculture Human excreta, lost Mil lio n to nn e s P pe r ye ar

(22)

largely contributed to the expansion of soybean cultivation in Cerrado biome in Brazil and further expansion is dependent on synthetic P fertilisers. On average, 28 kg P are added per hectareper year in Brazilian soybean production. The prognosis for a significant increase in soybean volumes by 2020 suggests an input of at least 0.8 Mton of P per year to meet the demand in 2020. This amount of fertiliser will then be at least 4.5% of the current global use of fertiliser P, which is a substantial amount for one crop in one country (Smit et al., 2009).

When cereals are used as livestock feed there is a substantial P-flow from the fertiliser input in crops to livestock production units. Besides the P-flow in cereals that are fed directly, feed co-products from cereals and pulses (e.g. wheat bran, rapeseed cake) also result in large P-flows into livestock production. For example, after milling wheat, the bulk of the P in the grain is in the co-product wheat bran, while the flour for human consumption has a low P content. After extraction of soybeans and rapeseed, most of the crop´s uptake of P is destined for the feed co-product meal/cake. The vegetable oil produced only holds a small amount of P. Production of pork and poultry animals in particular leads to large P-flows from mined fertiliser phosphates to crop production (grain and soybeans) via feed and food industry to concentrates used to feed the animals.

Since the main part of P in livestock feed ends up in manure, inadequate recycling of manure results in soil P accumulation in regions with high livestock density. In Denmark, where agriculture has a long tradition of animal husbandry, 1.4 ton of P per hectare of arable land was accumulated during the 20th century (Poulsen and Damgaard, 2005), i.e. around 14 kg P per hectare per year. In the Netherlands, where there has been a strong increase in intensive pig and poultry production since the 1950s, combined with imported feed, an accumulation of soil P at around 2 ton of P per hectare of arable land has occurred. In 2005, agricultural soils accumulated on average 20 kg of P per hectare in the Netherlands while the balance for agricultural soils in EU27 had an average surplus of around 8 kg of P per hectare (Smit et al., 2009). Continuous soil

P-accumulation poses an increased risk of P-leaching from agricultural soils (Heckrath et al., 1995) and thus an increased risk of eutrophication in future.

Land-related impacts

Land occupation

Agricultural land is divided into arable land and pasture and a significant share of this land is used for feeding the world´s livestock. By combining agricultural inventory data and satellite-derived land cover data, Ramankutty et al. (2008) estimate global arable land at 1,500 million hectares (Mha) and global pasture area at 2,800 Mha3 in the year 2000, totalling around 4,300 Mha. Steinfeld et al. (2006) calculate that around one third of total arable land is used to provide feed for livestock, including primary crops such as cereals and co-products, important protein meal from oilseeds processing, and also by-products from cereal processing and from citrus and sugar crops. This means that roughly 75% of all agricultural land today is used to produce feed for the world´s livestock, see Figure 4.

3_{Ramankutty et al estimate of global pasture is 18% lower than the standard FAOSTAT estimate of 3,440}

Mha. The major difference is found in Saudi Arabia, Australia, China and Mongolia and is likely due to the fact that pasture census data reported to FAOSTAT include grazed forestland and semi-arid land.

(23)

Figure 4. Share of global agricultural land (approximately 4,300 Mha) use for direct human consumption and for feed production to livestock (based on Ramankutty et al. 2008 and Steinfeld et al. 2006).

Figure 5 presents regional trends for land use in agriculture between 1991 and 2001. The expansion of agricultural land is greatest in Latin America and sub-Saharan Africa, mostly at the expense of forest cover. In Asia (mainly South Asia) and Oceania there is an expansion of prime arable land. Europe and North America show a net decrease in agricultural land use which has been a trend for the last four decades, coupled with stabilisation or increase in forest land (Steinfeld et al., 2006). High intensification of agriculture since the 1950s is one explanation for this trend.

Figure 5. Regional trends (annual growth rate, %) in land use for cropland and pasture between 1991 and 2001 (Steinfeld et al., 2006).

Expansion of livestock production is a key factor in deforestation, especially in South America, where around 70% of deforested land is occupied by pasture. Feed crops (mainly soybeans) cover a large part of the remainder (Steinfeld et al., 2006). The environmental impacts caused by deforestation are both local and global. Besides emissions of greenhouse gases due to deforestation, important carbon sinks are lost when forest cover is removed and replaced with agricultural crops. Rainforests in the Amazon region regulate the water balance and flow over the entire Amazon River system, influence the pattern of climate and air chemistry over much of the continent and may facilitate the spread of vector- and air-borne diseases in the region (Foley et al., 2007).

23%

12% 65%

Arable land, human food

Arable land, feed for livestock Permanent pasture, livestock -0,6 -0,4 -0,2 0 0,2 0,4 0,6 0,8 1

Dev Asia Oceania West Eur N Africa S-S Africa N America S America An n u al gr o w h r ate, %, 1991 -2001

(24)

Biodiversity

Land use, and land use change such as deforestation, impact greatly on biodiversity. Sala et al. (2000) examined change scenarios in global biodiversity for the year 2100 based on changes in atmospheric carbon dioxide, climate, vegetation, and land use and the known sensitivity of biodiversity to these changes. The study showed that for terrestrial ecosystems, land-use change is the most severe driver of changes in

biodiversity. Conversion of temperate grasslands into arable land, or tropical forests into pasture results in local extinction of most plant species and the animals associated with them, whose habitat is largely determined by the composition of plant species. In addition the Millennium Ecosystem Assessment lists land use change as the leading cause of biodiversity loss.

Rapidly growing demand for livestock products is the most important driving force for expansion of soybeans in South America (Steinfeld et al.., 2006). In Brazil (a major exporter of soy products to Europe) much of the growth in soybean production has taken place in the Cerrado biome where more than half of approximately 200 Mha has been transformed into pasture with monoculture grass species and cropland in the last 35 years. The Cerrado biome has the richest flora among the world‟s savannas (>7,000 species) and high levels of endemism (species not found elsewhere in the world). Deforestation rates have been higher in the Cerrado than in the Amazon rainforest, and conservation efforts have been modest: only 2.2% of its area is under legal protection. Numerous animal and plant species are threatened with extinction (Klink & Machado, 2005).

The intensification of agriculture since the 1950s has contributed to biodiversity decline and loss across Europe. Major effects from livestock production are due to emissions of reactive N, the most important being ammonia. A number of valuable European habitats have been shown to be seriously threatened by atmospheric N deposition, including freshwater, species-rich grassland and heathland (Leip et al., 2010) due to increased competition of N-adapted species at the expense of less adapted species.

Agricultural intensification has resulted in the homogenisation of large areas of Europe‟s rural landscape. Of special importance to livestock production is farming system specialisation (livestock versus arable) with the loss of mixed farming systems, and the creation of larger farming units leading to the removal of non-cropped areas and field boundaries (Leip et al., 2010). But there are also positive impacts from cattle farming; the biodiversity in European semi-natural grassland is very high and their management is dependent on grazing livestock. Very large proportions of Europe´s most threatened bird species, vascular plants and insects live in these grasslands and other High Nature Value farmland (Emanuelsson, 2008).

Soil fertility

Foley et al. (2005) estimate that due to increasing fertiliser use, some irrigated lands have become heavily salinized, causing the worldwide loss of 1.5 million hectares of arable land per year, along with an estimated $11 billion in lost production. Up to 40% of global croplands may also be experiencing some degree of soil erosion, reduced fertility or overgrazing. Pasture degradation related to overgrazing and lack of nutrient replacement is a major problem for many of the world´s grassland area (Steinfeld et al., 2006).

(25)

Greenhouse gas emissions

The FAO report Livestock´s Long Shadow (Steinfeld et al., 2006) initiated discussions and research around the role of animal products in climate change. It claims that global livestock production contributes to about 18% of global anthropogenic greenhouse gas (GHG) emissions, which was roughly 7.1 billion tonnes CO2-equivalent (CO2e) in early 2000 (Steinfeld et al., 2006). Emissions of fossil CO2 is of minor importance for the global livestock sector, rather CO2 releases from Land Use and Land Use Change (LULUC), mostly deforestation, are the overall dominating source of carbon emissions from the global livestock sector (see Figure 6). Feed digestion by enteric fermentation in ruminants is a major contributor to methane (CH4) emissions from livestock production but manure management is also important. Losses of nitrous oxide (N2O) from manure due to storage, spreading in fields (and thereafter soil-N- turnover) and indirectly also due to ammonia emissions from manure make up a substantial share of total GHG emissions (see Figure 6).

Figure 6. Emissions of greenhouse gases (CO2, CH4 and N2O) from the global livestock sector (Steinfeld et al.,

2006).

In a follow-up study, FAO estimates that the global dairy sector is responsible for approximately 3 – 5% (depending if including co-products beef) of global GHG emissions (Gerber et al., 2010).

Naturally there are considerable variations in how much the livestock sector contributes to total GHG emissions in different parts of the world. Leip et al. (2010) estimate that the GHG emissions from the livestock sector in the EU27 correspond to close to 13% of total EU emissions when also including LULUC. The land-related emissions are mainly in non-EU countries and a consequence of feed import to European livestock,

principally soymeal from South America. The main reason why Europe has a relatively lower share of total emissions from the livestock sector is due to high emissions in other sectors (e.g. transports, energy) and low emissions due to LULUC within Europe.

0 0,5 1 1,5 2 2,5 3 billion tonne s C O2 -e q u iv al e n ts N2O CH4 CO2

(26)

Carbon dioxide, CO2

As seen in Figure 6, emissions of fossil CO2 is of minor importance for the global livestock sector´s GHG emissions, but in a highly industrialised region such as the EU27, around 20% of total GHG emissions from the livestock sector come from fossil energy (Leip et al., 2010).

CO2 emissions from land-use change processes are closely connected to expanding agricultural production and land clearance. Carbon emissions from forest clearing constituted about one third of total anthropogenic CO2 emissions in the period 1850-2005. In the last 50 years, there has been a stabilising (or even decrease in agricultural land) in many regions but in the tropics, deforestation is still occurring rapidly. During the 1990s, it is estimated that tropical deforestation gave rise to CO2 emissions in the order of 1-2.2 GtC per year, comprising 14-25% of total anthropogenic emissions (Denman et al.., 2007). Today, land use change, mostly deforestation, accounts for around 10% of global CO2 emissions, and the emission trend has been falling over the last 10 years compared to levels during the 1990s (www.globalcarbonproject.org). The large uncertainties in estimates of GHG emissions from deforestation must be noted: release of black carbon is seldom included, nor is the loss of carbon sinks when forests are cleared.

Methane, CH4

As seen in Figure 6, enteric fermentation is the most important source of methane (CH4) emissions from livestock production. CH4 is also emitted from slurry storages and in a warm climate, this source can be substantial. Methane is around 25 times more effective in trapping heat in the Earth‟s atmosphere than CO2 and its atmospheric concentration has increased from around 715 ppb4 pre-industrial to 1,774 ppb in 2005, i.e. by close to 150% (Forster et al., 2007).

According to Steinfeld et al. (2006), global emissions from enteric fermentation and manure management in 2004 were 85.63 and 17.52 Mton CH4, respectively. This corresponds to approximately 2.58 billion tonnes CO2e which is around 5.3% of total GHG emissions in 2004.

The production of methane by ruminants normally contributes to the largest share of total GHG emissions. For European beef and milk products, methane represents around 45-50% when LUC-emissions are not included, but less than 40% of total emissions when LUC-emissions are included (Leip et al., 2010). Beef produced in South America, predominantly produced on grazing all year around, has generally higher CH4

emissions, contributing to as much as 75% of total emission (LUC not included) which is caused by high slaughtering ages and long calving intervals (Cederberg et al., 2011).

Nitrous oxide, N2O

As seen in Figure 6, manure storage, manure spreading in fields and ammonia emissions from manure give rise to a substantial share of nitrous oxide (N2O) emissions from the global livestock sector. Nitrous oxide is a potent greenhouse gas that is present in very low concentrations in the atmosphere. It is around 300 times more effective than CO2 in trapping heat and also has a very long atmospheric life-time (>100 years). Pre-industrial concentration was around 270 ppb N2O which has grown to 319 ppb in 2005, i.e. an 18% increase.

4_{ppb (part per billion) is the ratio of greenhouse gas molecules to the total number of molecules of dry}

(27)

Anthropogenic emissions of nitrous oxide (N2O) have been emphasised in the second half of the 20th century as a result of the strong increase in synthetic N fertiliser use in agriculture. Steinfeld et al. (2006) estimate that around 30% of the global livestock emissions of GHGs are made up of N2O (see Figure 6). Corresponding numbers for EU-27 N2O emissions from the production of milk, meat and eggs are around 23% of the sector´s total GHG emissions (Leip et al., 2010).

Chemicals

Pesticides

Pesticides are chemical substances or biological agents (such as bacteria) used to kill or control any pests, mainly weeds, insects and fungus. The pesticides used for plant protection in agriculture are herbicides (against weeds), insecticides (insects), fungicides (fungi), nematocides (nematodes) and rodenticides (vertebrate poisons) (Ongley, 1996). There are large economic incentives to use pesticides to reduce the occurrence of pests and hence increase the quality and quantity of crop yield. In addition, use of pesticides is less labour-intense than traditional agricultural

management practices and reduces the reliance on good crop rotation for pest control. However, the benefits of improved plant protection by using pesticides need to be contrasted with the disadvantages related to their potentially toxic effects on humans and ecosystems and on an increased risk of pests and weeds becoming resistant to them. The impact of pesticides on different organisms varies greatly depending on their toxicity (direct and indirect), persistence and fate. Some of the potential effects are death of organisms, disruption of hormonal systems, tumours, and cellular and DNA damages (Ongley, 1996). The use of pesticides and associated changes in management practices may also affect biodiversity and soil fertility (ibid). Extensive use and reliance on one or only a few pesticides may also result in increased resistance to the pesticide due to natural selection in the targeted weed-, fungi-, insect population. An urgent problem is a growing number of weed species that have evolved resistance to the important herbicide glyphosate and this is mainly occurring in areas where farmers grow crops (most importantly soybeans) that have been genetically engineered to tolerate glyphosate5 (Waltz, 2010; Cerdeira et al., 2011). Due to adaptation and natural selection in weed populations, the increased and often exclusive reliance on glyphosate to manage weeds in genetically engineered crop systems have led to that resistance to glyphosate having evolved in some weed species.

The impacts of pesticides in feed production are not sufficiently assessed in LCA studies of animal production. The problems lie in the great number of pesticides used, the variety of environmental effects (known and unknown) and the fate of the

pesticides. However, there is an on-going development of models and methods to be used in LCA to assess the environmental effects of pesticides (ENDURE, 2009). In some LCA studies, the inventory includes use of pesticides given as grams of active substances, but these numbers do not reflect the environmental effect of their use since toxicity and fate as well as dosages (i.e. grams of active substance per hectare) vary between pesticides.

There are limited detailed and up-to-date statistics on the global use of pesticides. The most recent statistics from the EU (1992 to 2003) show that fungicides and herbicides, measured as active substances, contribute to slightly less than 90% of plant protection

5_{Glyphosate (common commercial trade-name Roundup) is the most used herbicide in the world. It is a}

(28)

products used in EU15 and that five countries (France, Spain, Italy, Germany and UK) use 75% of products (Eurostat, 2007). The use of fungicides in EU15 has declined during the 1990s probably due to the replacement of products that are used in high dosages, e.g. sulphur used in, for example, vine. Generally, a high proportion of the products are applied on special crops like fruits and vegetables.

The FAOSTAT database provides statistics on consumption and trade of pesticides on a national level (FAOSTAT, 2011). However, consumption rates are not available in the database for large agricultural countries like the US, Canada, Australia and Brazil. The statistics on trade include import and export of pesticides, but do not show domestic consumption.

Medicines

Veterinary medicines are used to decrease the prevalence of disease and increase efficiency in modern animal production systems by, for example, preventing infectious disease, managing reproductive processes, and controlling parasites and non-infectious diseases (Sarmah et al., 2006).

Information on the amounts of veterinary medicines used is limited although for some, like veterinary antibiotics, these data are, to a certain extent, public in some

Scandinavian countries and the Netherlands (Sarmah et al., 2006). A recent study estimated the total mass of active substances in veterinary medicine in the EU at 6,051 tons, of which 5,393 tons were antibiotics (~90%), 194 tons anti-parasitics and 4.6 tons hormones (Kools, 2008). A review by Sarmah et al. (2006) summarised available data on the use of antibiotics worldwide. A total of 16,000 tons of antimicrobial compounds are used annually in US, about 11,200 tons for non-therapeutic purposes. In New

Zealand, around 53 tons of antibiotics annually are used in animal production systems, a large share of the country´s livestock is cattle fed on pasture which explains the

relatively small amount of antibiotics used despite a relatively large animal production system. Data on medicine use in developing countries are even less and estimations are rough and uncertain. Sarmah et al. (2006) report around 15 tons of active antimicrobials used yearly in animal food production in Kenya.

Antibiotics are, by far, the most commonly used veterinary medicines and of great importance to modern agriculture, not only to treat diseases but to promote growth and improve feed efficiency. Not all antibiotics administered are absorbed by the animal; sometimes 30-90% is excreted in manure. After excretion, the metabolites may still be bioactive and transformed back to the parent substance (Sarmah et al., 2006).

Veterinary medicines are distributed to the environment by a number of different

pathways. Medicines may be emitted from processing plants when produced or disposed of, whether unused or out-of-date. Medicines are spread to the environment after being excreted in manure. The medicines and metabolites are transported and distributed to air, water, soil or sediment on the basis of the physiochemical properties of various substances and the receiving environment (Boxall, 2003).

Boxall et al. (2006) analysed the prevalence of ten veterinary medicine substances in soil and the potential of these substances to be taken up by vegetables. The results indicated that certain measurable residues were likely to occur at least five months after having entered the soil (mostly via livestock manure). The study also showed uptake of four substances in carrots and three substances in lettuce leaves. All detectable levels were below the so called acceptable daily intake (ADI) values, indicating low risk for affecting human health. The results did however show a need for caution when

(29)

considering long-term prevalence in the environment of veterinary medicines with low ADI (Boxall et al., 2006).

Studies have evaluated the toxicity of veterinary medicines in organisms such as fish, daphnids, algae, microbes, earthworms, plants, and sometimes dung invertebrates. With some exceptions, the environmental concentrations of veterinary medicines are

generally below effective concentrations in standard organisms. The risk of acute environmental effects is thereby low for most veterinary medicines used. The effect of pharmaceuticals on the environment has however emerged as a scientific field only since 2000. It is mostly single compound effects that have been examined; the pathways of degradation products and how these affect the environment have been less studied. By neglecting the metabolites, some conventional eco-toxicological assessment methods may underestimate the overall environmental effects of compounds. In addition to this, several veterinary medicines are sometimes used at the same time during treatments and the possible synergic effects of the different compounds in mixtures are difficult to examine accurately. These effects include synergism, additivity and antagonism which can decrease or increase the compounds´ effect on the

environment. The environmental effects from long term and low dosage exposure to various veterinary medicines also need further research (Boxall, 2003).

The main concern regarding the widespread use of antibiotics in veterinary medicine is the development of resistant bacterial strains, a health risk to both humans and livestock animals. Of special concern is antibiotic therapy in food-producing animals. Direct contact with animals, and thereby the risk of bacteria spreading via the food chain, enables the selection of bacterial strains resistant to antibiotics used in human therapy. Concerns regarding the risks of bacterial resistance led to the European ban on

antibiotic growth promoters issued in 2006 (Kemper, 2008). In the US, a national programme (NARMS) monitors the occurrence and distribution of antibiotic resistant bacteria in food (Sarmah et al., 2006).

(30)

(31)

5. METHODS FOR ASSESSING ENVIRONMENTAL IMPACT

Life Cycle Assessment (LCA)

Life Cycle Assessment (LCA) is a tool for assessing the environmental impact of a product or service. The basic principle of LCA is to follow the product throughout its entire life cycle. The product system studied is delimited from the surrounding environment by a system boundary. The energy- and material flows crossing the boundaries are accounted for as input-related (e.g. resources) and output-related (e.g. emissions to air), see Figure 7. LCA methodology is often used in well-defined parts of a product´s life cycle. For animal products, primary production, i.e. activities until the farm-gate, represents the dominating part of total environmental impacts. Therefore, many LCAs of livestock production only include the farming system and the input of resources to this system.

RESOURCES (Raw) materials Energy

Land

EMISSIONS

Emissions to air and water Waste

The LCA methodology has been developed within a general framework established by SETAC6 and further harmonised in an international standardisation process (ISO)7, see further ISO 14040 (ISO, 2006a) and 14044 (ISO, 2006b). An LCA consists of four phases. In the first phase, the definition of goal and scope, the aim and the range of the study is defined. Important decisions are made concerning the definition of the

functional unit (i.e. the reference unit), choice of allocation methods and system

boundaries. In the inventory analysis, information and data about the system is collected and relevant inputs and outputs are identified and quantified. In the impact assessment, the data and information from the inventory analysis are linked with specific

environmental parameters so that the significance of these potential impacts can be assessed. An example of this procedure is the conversion of greenhouse gas emissions into CO2-equivalents. In the final interpretation phase, the findings of the inventory analysis and the impact assessment are combined and interpreted to meet the previously defined goals of the study.

6_{SETAC is short for the Society of Environmental Toxicology and Chemistry and this organisation is one}

of the driving forces in the standardisation of LCA methodology.

7

ISO is short for the International Organisation for Standardisation. Raw material extraction xtraction Processing Transportation Manufacturing Use Waste treatment

(32)

LCA methodology applied to animal production

There are some methodological difficulties that arise in LCAs of animal production. For example, decisions on how to set the system boundaries and define the functional unit and what method to use for handling by-products are crucial and can greatly influence the outcome of an LCA. One important aspect of animal production systems is that the degree of environmental impact is not only a function of resource use and emissions per product unit, but is also related to the spatial concentration; this is important when assessing nitrate and ammonia losses. Finally, impact assessment of land use is still undergoing methodological development and the need for consensus in this area has been emphasised during recent years due to growing interest in carbon footprint analysis (see next section for more details). In addition biodiversity impacts lack a

methodological consensus. System boundaries

The farmed soil is generally regarded as a part of the technical system in LCA of agricultural products and this is a logical and reasonable presumption since soil is the precondition for all cultivation. However, the time horizon must be well defined when nutrient supplies and flows are described, since soil has a capacity to store nutrients for long periods of time. Normally, the chosen time period in an LCA is one agricultural year since the statistics for grain and milk yield or leaching data are often reported annually. The soil thus leaves the system at the end of the time studied (i.e. crosses the time boundary) but differences in the contents of nutrients in the soil should not be neglected. A very common feature of intensive livestock production systems is soil P accumulation as a consequence of surplus manure which leads to a risk of P losses into the environment in the future. Consequently there is a time-lag between the input of excess P into the soil and its possible future contribution to eutrophication. A nutrient balance can add useful information to an LCA and indicate potential losses in the future. Also when calculating soil carbon balances as an effect of different feed crops and management, the chosen time period is very important (see next section for more detail).

Functional unit

The definition of the functional unit (FU) is very important in LCA and it should be well defined and provide a strict measure of the function that the system delivers. Since it is the base for analysis it must be relevant and it must be possible to relate all data used to the functional unit.

LCAs of milk production often use fat- and energy corrected milk as the FU. This is a straight forward base for calculation since data on fat and energy content in milk is normally included in production statistics. In LCAs of meat production, there is a larger variation in how the FU is defined which can be confusing when interpreting the results. Most commonly in meat LCAs, the FU is kilograms of live-weight gain, carcass weight (slaughter weight, meat with bone) and bone-free meat weight which is the final product for human consumption. Generally, production statistics are given as carcass weight, often along with consumption data. But the yield of bone-free meat from the carcass differs between different livestock species and when comparing results from different meat products it is necessary to have a clear understanding of how the FU is defined in the study. When comparing different LCAs of meat, it is important to interpret the FU correctly and recalculate it when needed to be able to make a fair comparison.

Sometimes, in LCA of agricultural products, hectare of farmed land is used as a

functional unit in combination with the product from the land. The reason for this is that agricultural land not only produces food but can also provide other benefits, such as biodiversity and a varying landscape. An example is grazing cattle on semi-natural

(33)

grassland which has a positive environmental impact acknowledged in many agri-environmental programs in the EU, since a very large proportion of Europe´s most threatened bird species, vascular plants and insects live in these areas. In some cases, environmental subsidies for such meat production can provide a substantial income to the farmer in addition to payment from the beef market. In this case, ecosystem

services, such as biodiversity connected to grazing cattle, can be said to be a by-product (or more correct a “by-service”) of beef production. However, the main methodological problem of including land in the functional unit is that land is a necessary and input-related resource in production and this is not made obvious when land is included in the FU. Instead one can form the idea that land is an unlimited resource, which obviously is not the case.

Inventory analysis

One problem connected with the inventory analysis in LCA of animal production is associated with production data used in calculations (e.g. average milk yield, fertiliser rate, feed intake, etc) which are often collected as mean values from statistics or data from farms. There are often considerable uncertainties in production data due to i) inadequate statistics and ii) differences in management practices among farms resulting in variations in production data across farms. Henriksson et al. (2011) calculated the national average of GHG emissions for Swedish milk at 1.12 kg CO2e per kg energy corrected milk (ECM) and then analysed the variation in this average estimate caused by differences in production parameters between dairy farms. The variation in the GHG emission per kg average for Swedish milk was found to be ±17% due to variation in management practices and milk yield between farms.

The variation in one important production parameter is shown in Figure 8 - feed efficiency (milk production per kg feed intake). Based on data from 1,050 dairy farms participating in a feeding advisory service in Sweden, there is an average production of 1.44 kg ECM per kg dry matter feed with a variation between 0.96 – 1.82 kg ECM per kg dry matter feed (Henriksson et al.., 2011). The large variation in production and management methods between farms means that data needs to be collected from quite a large number of farms when performing LCA of animal products aiming to compare different systems.

LCAs of animal production are sometimes based on models on nutrient requirements and metabolism of animals when estimating feed intake which is a very important production parameter. Using models, it is difficult to reflect the complexity of a

livestock production system and the large variations between farms. Statistical analysis needs to be included in LCAs to show better the uncertainty range in results but this is rarely done.